Crystallization of struvite in the presence of calcium ions: Change in reaction rate, morphology and chemical composition

Abstract Ammonium, phosphorus, and potassium from wastewater treatment with a coexisting ion of calcium may be recovered simultaneously through struvite and struvite-(K) crystallization. This paper presents the quantitative assessment of the impact of calcium ions on the kinetics and crystallization of those crystals. Initial solutions containing dose levels of Ca2+ ion and pH 9 were set up for experiments in a stirred laboratory crystallization at ambient temperature. According to the pH reduction data, the observed precipitation kinetics followed in two steps; the first step (0–3 min) and the second step (3–60 min) in which linear regression analysis of both kinetic data fit with first-order rate constants. In the absence of calcium, the computed kinetic constants are respectively 2.568 h−1 for the first stage and 1.548 h−1 for the second stage. The kinetic rate constants followed with the increased dose of Ca (Ca/Mg > 0.5), which lengthened the crystallization of multiphase crystallization of Mg and Ca-phosphates. Accordingly, calcium had a negative effect on the morphology, purity, and quantity of the final crystalline product. This quantitative understanding of how calcium affects the crystallization of struvite and struvite-(K) reliably improves knowledge about controlling the quantity of wastewater recovery products.

D.S. Perwitasari received Ph.D. in Mechanical Engineering from Diponegoro University, Indonesia. The focus of her research is currently to conduct the synthesis of struvite through aqueous crystallization for the waste streams. S. Muryanto is a Professor in the Chemical Engineering at the Department of Chemical Engineering, UNTAG University in Semarang, Indonesia. The focus of his research is to design and engineer struvite crystallization for the waste streams. J. Jamari is a Professor in Mechanical Engineering, at Diponegoro University. Research interest covers many aspects of material design, surface treatment, including wear characterization and lubrication technology. A. P. Bayuseno is a Professor in Material Science and Engineering at Diponegoro University, Indonesia. Research interest covers many aspects of ceramics design, applied crystallography, including materials characterization and waste processing.

PUBLIC INTEREST STATEMENT
This research paper provides insight into the biomineralization of Mg-and Ca-phosphates in wastewater streams. As compared to the Caphosphate-bearing minerals, struvite [MgNH 4 PO 4 •6H 2 O] formed during treatment of wastewater, provide the benefits of recovering N and P simultaneously by yielding fertilizer of low solubility with the slower-released rate of nutrients than other fertilizer product. However, struvite formation can be inhibited by the presence of Ca 2+ in the solution. Our work proposed the unified approach that deals with a strategy for synthesizing struvite and struvite-(K) from the synthetic wastewater with and without calcium, and the quantitative interpretation of biomineralization products in the synthetic one. Following Xray, and SEM analysis, it was found that struvite and struvite-(K) were highly inhibited and present in the microspheres when Ca-phosphate minerals were developed.

Introduction
Over the last decades, the industrial demand for phosphorus (P) for fertilizer production has increased dramatically along with the increase in the worldwide population, while natural phosphorus-bearing minerals are depleting rapidly (Scholz et al., 2013). On the other hand, the aquatic system in many regions worldwide is surplus insoluble phosphorus due to gray water from household liquid discharge impregnating with phosphate detergents. Such P-rich domestic effluents, along with agricultural run-off where fertilizer may have been used extensively, instigate the uncontrolled growth of algae and other aquatic plants. The unrestrained growth water florae, termed eutrophication, is unsightly as well as detrimental to environments . Detergents containing nitrate or phosphate, fertilizer in wastewater may be found in many regions, where these substances are often discharged into an aquatic system as a result of inadequate wastewater treatment resulting in eutrophication. Therefore, extensive research to alleviate such environmental burden on the environment utilizing different techniques has been carried out.
Since phosphorus (P) is one of the vital elements required for life (Ali & Schneider, 2005;Lahav et al., 2013), the techniques capable of getting rid of eutrophication coupled with recovering the P for subsequent use are obviously more preferable, and among the various methods applied, those which could be with the intention to recover the nutrient. It is believed that simultaneous crystallization from a solution of the common three species in wastewater: Mg, N, P, termed MAP solution, yielding crystalline struvite [MgNH 4 PO 4 •6H 2 O] and its variant, e.g., struvite(K) [KMg(PO 4 )•6(H 2 O)] is promising and worth implementing (Bhuiyan, Mavinic, Koch et al., 2008a;Bouropoulos & Koutsoukos, 2000;Doyle & Parsons, 2002). This crystallization method can be used for the concurrent reclamation of ammonium, potassium, and phosphorus from the wastewater (Ali & Schneider, 2005). Accordingly, it has been proposed as a simple and realistic means of providing a long-term supply of phosphorus (Lahav et al., 2013).
Recently, much attention has been given to implementing crystallization of struvite and/or struvite-(K) for ammonium and phosphate recovery in various types of wastewater (Bhuiyan, Mavinic, Koch et al., 2008a;Bouropoulos & Koutsoukos, 2000;Doyle & Parsons, 2002), and in the hydrothermal solution (Bayuseno & Schmahl, 2018. For this reason, the crystallization of struvite and/or struvite-(K) in MAP solution (magnesium, ammonium, and phosphate) containing potassium may be performed by adding magnesium ions in a simple bath reactor. As reported, the crystallization products of struvite and struvite-(K), both are highly insoluble in water, can be used as fertilizer (Le Corre et al., 2009).
Further, the struvite crystallization from an MAP solution is commonly represented by the following equation (Doyle & Parsons, 2002): Hence, the mineralization of struvite depends mainly on pH, and n may vary between 0, 1, and 2. Also, struvite could be crystallized from the supersaturated solution depending on the MAP molar ratio in which a molar ratio of 1:1:1 could be a favorable condition of its formation. Fundamentally, the mechanism of the crystallization of struvite is governed by a number of factors including nucleation and crystal growth, which can be controlled by physicochemical factors such as pH, mixing intensity, temperature, and impurities in the solution (Li et al., 2016;Y. Song et al., 2014). In addition, struvite crystallized from the supersaturated solution reflects the ion activity product (IAP) of Mg 2+, NH 4 + , and PO 4 3-, which has a value above its solubility product (KSP; Bayuseno & Schmahl, 2018;Doyle & Parsons, 2002). Specifically, the successful crystallization of struvite may be under the control of pH solution and supplied by phosphate ions. Thus, the most recent study on struvite crystallization focused mainly on controlling the pH solution with the variable molar ratio Mg/P in solution chemistry Shih et al., 2017;Wang et al., 2005). As previously indicated, the pH range of 8-9 and the Mg/P ratio of 1.0-1.5 have been shown to be favorable wastewater conditions for the recovery of phosphate ions (species of PO 4 3and their variants dictated by the pH levels) in the form of struvite (Bhuiyan et al., 2007;Bhuiyan, Mavinic, Koch et al., 2008b;Doyle & Parsons, 2002).
Physicochemical factors affecting struvite crystallization may also relate to the chemical composition of a digestion solution, which comes from the different sources of wastewaters with dissolved foreign ions (Rahman et al., 2011,Yan & Shih, 2016. In some cases, other constituents of wastewater could be discharged from wastewater treatment plants that subsequently react with MAP ions leading to crystallization of struvite (Le Corre et al., 2009;Karabegovic et al., 2013). In some cases, different sources of wastewater clearly contain different dissolved ions (Rahman et al., 2011,Yan & Shih, 2016; hence dictating various physicochemical variables for successful crystallization of struvite (Le Corre et al., 2009;Karabegovic et al., 2013).
It is common to estimate the purity of struvite as the ratio of NH 4 /PO 4 in the precipitates. In regard this estimation, it has long been observed that a clear trend in the decreasing of struvite purity was seen when higher Ca/Mg ratios were employed (Wang et al., 2005). Using synthetic wastewater mimicking the effluents of anaerobic digestion lagoons treating piggery wastes, Wang et al. (2005) reported a gradual decrease in struvite purity from 85% down to a mere 38% when the Ca/Mg were increased four times, i.e. from 0.5:2 to 2:2. With such a decrease it can be assumed that Ca is the main interfering cation during the precipitation of struvite capable of replacing the NH 4 ions in the process.
Compared with calcium phosphate crystallization, MAP one involves the process of recovering ammonium and phosphate on which there is no need for induction of seed crystals. The crystallization method is also simply used to recover phosphates (Liu et al., 2013;song, Donnert et al., 2007;Song, Yuan et al., 2007). MAP precipitation can also be applied as a slow-release, long-acting fertilizer (Liu et al., 2013). Because of its favorable technique, MAP crystallization could be used for phosphate recovery from pig sewage (Capdevielle et al., 2013;Crutchik & Garrido, 2011).
Conversely, high concentrations of Ca 2+ in the MAP solution could increase phosphorus removal efficiency from 58% to 92% (Huang & Liu, 2014;Lee et al., 2013;Pastor et al., 2008). This substantial gain in the phosphorus removal efficiency occurred favorably at Ca/Mg molar ratios above 2, while ratios below 0.5 would not impact the crystallization of struvite (Desmidt et al., 2013). The impact of calcium on struvite crystallization was also observed at low concentrations of magnesium and ammonium in the solution, resulting in the formation of other undesirable minerals and a change in struvite morphology (Le Corre et al., 2005).
Currently, the impact of varying Ca/Mg ratio on the struvite crystallization in the wastewater has become an intensive study focusing on morphology and purity of the product, whilst the use of simple reactors with the batch system for phosphate recovery by adding magnesium reagent on the wastewater is commonly implemented (Dalecha et al., 2014). In particular, batch crystallization offers numerous advantages in the laboratory and industrial applications. The laboratory crystallizer also assists in characterizing the crystallization kinetics and crystal size distribution (RDS) and determining the impacts of process conditions on these kinetics and CSDs. Therefore, wastewater treatment in the batch system is seen as a good deal more economical with low production capacities of approximately 1 m 3 of product per day or less. In addition, batch crystallization offers an advantage with the ability to produce uniform particle size.
However, the quantification of precipitates containing struvite and struvite-(K) from a batch crystallization process as indicators of economic viability has not been specifically investigated. The corresponding data of the mineral compositions of solid products precipitated from the aqueous solution at varying Ca/Mg ratios and their effects of competition between Ca-P and Mg-P, including their crystallization kinetics, were nevertheless limited. Consequently, this batch crystallization study with synthetic wastewater was designed to study the influence of Ca2+ ions on the crystallization of precipitates based on the analysis of the analytical tool. In addition, the Mg and Ca-competitions for the recovery of phosphates from synthetic wastewater as potential phosphate minerals were studied using XRPD (X-ray powder diffraction) and SEM (scanning electron microscopy) analyses combined with kinetic data. The predicted limit value of the Ca/Mg molar ratio controlling crystallization in the synthetic wastewater would be helpful in managing the purity and quality of struvite and struvite-(K) in the future.

Batch precipitation experiments
The struvite crystallization experiments were carried out using an agitated laboratory glass beaker of 200 ml volume. Anhydrous magnesium chloride (MgCl 2 ) and ammonium dihydrogen phosphate (NH 4 H 2 PO 4 ) crystals (Merck, AR grades) were used for the struvite crystallization providing ions: Mg 2 + , NH 4+ , and PO 4 3− necessary for the reaction. Moreover, a stock solution concentration of 0.17 M each was prepared from those chloride and phosphate powder crystals, which were analytically weighed (WANT Balance, FA-N Series, 0.0001 g) and then diluted separately with distilled water.
The dissolution was carried out using standard laboratory glassware (graduated cylinders, volumetric flasks, and other glass apparatus of various sizes). In view of the pH solution being affected by absorbing CO 2 from the air into the solution due to long-standing distilled water, boiling the distilled water was necessary and left to cool in a closed container prior to the dissolution. Next, a pH of 9.00 in the mixed solution was set up by drop-wise addition of 0.2 N KOH (Bhuiyan et al., 2007;Bhuiyan, Mavinic, Koch et al., 2008b;Doyle & Parsons, 2002). The pH adjustment was required for struvite crystallization that occurs only in basic conditions. In the experiment, the chemical compositions of crystal-forming solutions are presented in Table 1.
In the individual crystallization run, 175.5 mM MgCl 2 and 175.5 mM NH 4 H 2 PO 4 solutions were dissolved in 50 ml each of a 200 ml-glass beaker. The beaker containing mixed solution was constantly stirred to ensure that the crystallizing solution was homogeneous. An impeller speed of 200 rpm was found to be appropriate for the homogeneity of the solution without breaking the crystals. The pH solution was continuously read using a pH meter (METTLER Toledo-portable pH meter), which was immersed in the mix solution for about 80 min. Eventually, the precipitates were quickly filtered by filter membrane (Whatman ® -WHA1001325-grade 1) and air-dried in a secure place. The precipitating solids were kept for subsequent material characterizations, i.e. scanning electron microscope (SEM) coupled with energy dispersive spectroscopy (EDS), and X-ray powder diffraction (XRPD) method.

Kinetic analysis
As stated previously, the current crystallization was focused on the influence of impurity on crystallization, i.e. calcium ions since Ca 2+ is one of the major cations present in wastewater. It was previously reported that the presence of impurities, i.e. even in ppm quantities, might influence significantly on the crystallization Muryanto & Bayuseno, 2014). For this experiment, powder crystals of CaCl 2 •2H 2 O were dissolved in distilled water and subsequently diluted into the MgCl 2 solution in predetermined concentrations (1, 10, 20 ppm and, Ca: Mg ratios of 0.5, 1 and 2). The experiment was carried out at ambient temperature to make sure that no ammonia/nitrogen component evaporated from the solution. In fact, many WWTP (wastewater treatment plants) normally work at ambient conditions (Shih et al., 2017). Correspondingly, experimental runs were performed to predict the length of the crystallization run and to validate the reproducibility of the measurements. Therefore, it was then decided that the run was done in triplicate for each measurement.
In the present study, the struvite growth rate was estimated by observing the pH changes as the crystallization progressed; while the pH drop corresponded to the decrease in [Mg 2+ ] as described in Equation (2). As shown in Equation (1; see Introduction), the rate of struvite crystallization can be presented as either the rate of reduction of [Mg 2+ ] or the rate of increase in hydrogen ion concentration, [H + ]. Mathematically, the relationship is expressed (Muryanto & Bayuseno, 2014) as: where: In batch experiments, a digital pH meter recorded the pH continuously during precipitation. In the kinetic analysis, a pH solution was seen over time Muryanto & Bayuseno, 2014). After 80 minutes of testing, the stirrer was turned off, then the solution was separated by 0.45 µm filters. Eventually, the precipitates were cleaned with deionized water to remove impurities from the crystal surfaces. In this study, only chlorides and alkali ions could be released during scrubbing. The precipitate slurries were then dried at room temperature for 48 hours, and stored in a small plastic container for later characterization.

Mineralogical phase characterization
The obtained precipitating solid was dried and then subjected to the X-ray powder diffraction (XRPD) measurements. Initially, the dried solid was ground to provide a grain size of less than 75 µm and later mounted in the Aluminum-XRPD sample holder. The XRPD data were collected by Bragg Brentano (Philips 1830/40) X-ray diffractometer at room temperature using Cu Ka radiation, where the fixed measurement parameters (5-85 O 2θ; 0.02 O steps; 15s/steps) were employed. The identified phases in the samples were acquired by the diffraction line-matching program (Match software), whilst the resulting phases were then verified by the qualitative and quantitative Rietveld method (Fullproof-2k, version 3.30;Rodriguez-Carvajal, 2005;Wiles & Young, 1981;Young, 1993). The abundant crystalline phase (weight. %) in the sample was calculated using refined results of the unit cell parameters and the scale factor (Hill & Howard, 1987). Detailed Rietveld refining procedures were established according to the methodology reported previously (Bayuseno & Schmahl, 2015;Mahieux et al., 2010).
Further, the ground powder samples were mounted with glue on the Al-sample holder surface, and coated with carbon for the morphological analysis by SEM (JEOL JSM 5200), while the composition of the chemical element was determined by the EDX system.

Precipitation kinetics
The precipitation kinetics was calculated based on the reduction of the Mg 2+ in the solution by manipulating the decrease in pH values as shown by the pH meter readings (Muryanto & Bayuseno, 2014). The manipulation was fairly simple in that during the process of the precipitation of struvite it was demonstrated that [Mg 2+ ] = [H + ] throughout in the study Nelson et al., 2003;Quintana et al., 2005). In fact, struvite precipitation is mainly under control by pH, initial relative MAP concentrations, and other coexisting cation of Ca 2+ . Accordingly, pH is regarded as a key aspect to control struvite crystallization (shape, morphology, and purity). Moreover, the present kinetic experiments were relied on the change of pH that could be related to the decreased Mg 2+ concentration at the ambient temperature and at a constant stirring speed of 200 rpm with varying Ca-concentrations. Figure 1 shows a drop in pH over time in the solution with and without variable calcium concentrations at 30 O C. In the absence of Ca 2+ , the pH falls abruptly over the first three (3) minutes and then gradually decreases (Figure 1a-b). Just after 80 minutes, the pH solution appears to be fairly stable. During the first 3-minute period, the pH decrease corresponded to the struvite crystallization (Ali & Schneider, 2006;Darwish et al., 2017;Kofina et al., 2007;Prywer et al., 2012;Rahaman et al., 2008), whereas the second phase may concern the formation of other Mg-bearing minerals. The response rate constants were then predicted by linearly fitting the experimental data in the modified first-order kinetic model ( Figure 2). The calculated rate constants are presented in Table 2. In the first period, this model provided an estimated rate constant of 2.568, 4.074, 6.684, 15.33 h −1 at Ca 2+ concentrations of 0, 1, 10 and 20 ppm, respectively. Moreover, those results in the absence of Ca 2+ are in reasonable agreement with those reported Nelson et al. (2003) and Rahaman et al. (2008).
In contrast, the increasing reactions to the presence of Ca were observed, in turn, inhibiting the struvite crystallization (Figures 3a, b). Here, varying rate constants (5.676, 3.060, and 4.392 h −1 ) with respect to Ca concentration were obtained in Ca/Mg ratios of 0.5, 1, and 2, respectively (Table 2). In this regard, variable-rate constants may relate the constituent ion concentrations controlling to the supersaturation ratios of precipitated phosphate-bearing minerals. Evidently, calcium delayed the induction time of struvite crystallization for the first appearance of its crystal and interfered with the growth rate of struvite (Bouropoulos & Koutsoukos, 2000;Le Corre et al., 2005).
Regarding the accuracy for the kinetic model, the present developed model was compared with other models reported for struvite precipitation from wastewater stream in which some of the published papers had been refereed (Ali & Schneider, 2006;Le Corre, 2006;Darwish et Kofina et al., 2007;Nelson et al., 2003;Prywer et al., 2012). Evidently, the obtained uncertainties of parameters and significant numbers in this current study were in good agreement with those reported in those papers. Accordingly, the present kinetic analysis provided accurate results that agreed very well with those of published papers for struvite crystallization in the batch system.

Purity and morphological crystal products
The crystallized products without and with calcium (0, 1, 10, and 20 ppm) were assessed qualitatively by a computerized search-match procedure of the XRPD method, relying on the powder diffraction file (PDF) from the International Centre for Diffraction Data (ICDD). As a result, most of the struvite-(K) (PDF#70-2345) and struvite (PDF#71-2089) were found, along with sylvite (PDF #73-0380). These phases identified were then validated by the Rietveld refining method as shown that the observed (Y ObS ) profile patterns agreed very well with calculated (Y calc ) profile ones (Figure 4a for crystal samples collected with 20 ppm Ca 2+ ). Moreover, three phases of struvite, struvite-(K), and sylvite were confirmed to be found in the precipitation collected with increases in Ca concentrations (0 to 20 ppm) ( Figure 4b). Clearly, the composition of the collected precipitate phases was not modified under the influence of these Ca 2+ concentrations. As a result of the minerals formed a solid solution, precipitates can be made from a  mixture of any single crystal. As shown previously (Figures 1 and 2), two stages of precipitation are suggested to occur in the solution (Ca/Mg ratio > 0.5) at the temperature of 30 O C, upon which two Mg-bearing minerals would be formed during crystallization. In this regard, the pH decrease was observed in two distinct steps; in the first stage (0-3 min), the pH decreased quite sharply; but in the second stage, the pH decreased progressively (3-60 min). These steps were supported by XRPD results confirming that two crystalline phases were formed. Correspondingly, struvite-(K) was precipitated instead of struvite according to (Equation 3): Hypothetically, struvite-(K) was formed because of excess KOH ions present when adjusting the pH in the MAP solution. An initial pH-9 solution with a Ca/Mg ratio higher than 0.5 was favorable for the recovery of phosphate and potassium as struvite-(K; Bouropoulos & Koutsoukos, 2000;Song, Donnert et al., 2007).
Additionally, XRPD diffractograms of precipitating solids at Ca/Mg ratios are shown in Figure 5. Each peak profile has been checked using the Rietveld refining method and linked to the standard mineral database. With Ca/Mg ratios of 1 and 2, the formation of struvite, struvite-(K), sylvite, hydroxyapatite (Ca 5 (PO 4 ) 3 OH), dolomite (CaMg(CO 3 ) 2 ), portlandite (CaOH), Mg-whitlockite (Ca 9 Mg (PO 4 ) 6 PO 3 OH), newberyite (MgPO 3 OH)•3(H 2 O) could be affirmed. It may also include amorphous calcium precipitate, which could not be fully identified with the XRPD method. However, this suggestion would be later justified by the EDX analysis. When the chemistry solution has a molar ratio of Ca:Mg > 0.5:1, there is evidence of the formation of hydroxyapatite, dolomite, and portlandite associated with the presence of Ca 2+ ions, which competes with phosphate ions in crystallization processes (Le Corre et al., 2005). As a result, the interaction of calcium with phosphate ions in wastewater systems can typically produce poorly crystallized hydroxylapatite based on the following reaction (Equation 4): Instead, under the condition of a (Ca/Mg) ratio greater than 0.5, the presence of calcium was reported to insignificant contribution to the P removal efficiency but only influence the purity of struvite crystals as confirmed by the XRPD analysis. In some cases, the struvite-like crystals might be enclosed with an amorphous phase, probably amorphous hydroxylapatite (Le Corre et al., 2005).
Other calcium effects on the development of crystal morphology were also examined with SEM-EDX. In the absence of calcium, struvite crystals with needle morphology (40 µm long and 10 µm wide) were evident in the SEM images ( Figure 6a). Apparently, the growth of the struvite crystal appeared to extend along the longitudinal axis. Additionally, struvite-(K) with an elongated rectangular bar-shaped morphology could be observed. The corresponding EDX spectrum confirmed the distinct peaks associated with K + , Mg 2+ , O 2-, and P 3− (in % mass), for constituent minerals of struvite and struvite-(K). It was noted here that the carbon peak could be observed to relate the carbon-coated on the surface of the sample for the analysis. Instead, at a low calcium concentration (10 and 20 ppm), the crystals of the struvite and struvite-(K) are shown to be attached with a precipitate (Figure 6b-c), suggesting that no other phase precipitates on those crystal surfaces interacted with their crystallization and the less amount of Ca 2+ could be absorbed on the surface of those crystals.
The EDX mapping analysis was afterward aimed at a specific area localized on the SEM micrographs of specimens with a coexisting cation of Ca 2+ for the molar ratio Ca/Mg more than about 0.5. Their spectra and the specific distribution mapping of elements were presented in Figures 7a, b and c, respectively. Under the Ca/Mg ratio of 1/2 or above, as confirmed from SEM-EDX analysis, other phases might be precipitated other than struvite and struvite-(K) crystals, while cation of Ca 2 + made the interaction with their formation through the absorption of Ca 2+ on the surface of both crystals. This condition proposed that the evolution of the struvite and struvite-(K) crystals appeared to be inhibited and would have formed an amorphous material. For samples with an Ca/Mg ratio of 1 or greater (Figure 7b-c), EDX analysis also supported earlier XRPD findings confirming that precipitates formed in this work may include both amorphous and crystalline calcium phosphate. Additionally, the amorphous phase formed on the precipitating solid could be deduced from the multitude of background noise in the XRPD diagrams. In Figure 5 for the sample with the Ca/Mg ratio > 0.5, the XRPD diffractogram displays the background noise, while the struvite peaks are always identifiable. Likewise, the XRPD pattern for the samples is illustrated in Figure 5, which has provided evidence of decreased purity of struvite with the evolution of Ca phosphate minerals into amorphous or crystalline forms (Le Corre et al., 2005).
According to SEM observations (Figure 7a), some needle-like morphology for struvite still remained, but most were aggregated with irregular crystal sizes and shapes, which are similar findings reported previously (Le Corre et al., 2005;Li et al., 2016). Likewise, the sample with the ratio Mg: Ca of 1:1 has some finer crystals associated with the struvite crystals observed. However, in the sample at the Ca/Mg ratio of 2:1, the needle morphology had disappeared and had been replaced by an irregular morphology (Figure 7c). This can be a result of multiphase precipitation as shown in Table 3. Here, the EDX analysis also supported the formation of Ca-phosphate minerals with higher calcium content, since both EDX spectra have higher peaks of P, Ca, and O than Mg (Figure 7b, c).
Apparently, the purity of struvite and struvite-(K) was significantly controlled by the higher ratio of Ca/Mg 1:2 (Figure 6), including the ammonium content in the chemistry solution (Le Corre et al., 2005). Although Ca/Mg ratios were similarly applied in the previous study (Le Corre et al., 2005), the present study employed a concentration of ammonium in 1:1:1 for the MAP molar ratio, which is more typical of dairy wastewater (Guillen-Jimenez et al., 2000). Where ammonium and phosphate are at the same concentration in the solution, the morphology of the product can be less sensitive to changes in Ca content. However, the degree of supersaturation of magnesium, ammonium, and phosphate in struvite production may more than offset the cost of chemical treatment to reduce the Ca content, making it less economically feasible. Accordingly, one of the main areas of reflection for reducing the concentration of it becomes a consideration for further research.
Further, different treatments of wastewater containing high calcium have been proposed in terms of their basic principles, applications, costs, maintenance, and suitability (Huchzermeier & Tao, 2012). Currently, the precipitation method has been a proven technology for wastewater with high calcium content turning calcium carbonate. Calcium carbonate precipitation has been reported to treat wastewater in anaerobic reactors, providing a cost-effective method, easy implementation, and high efficiency (Xiaoning Liu & Wang, 2019). Hence, a degree of compromise should be considered for pre-treatment by calcium carbonate precipitation, because there is a competition between quality benefits of calcium carbonate precipitation and chemically enhanced wastewater treatment for subsequent P recovery efficiency and solution pH. In this way, the proposed treatments for wastewater with high calcium may combine two methods, in which calcium carbonate precipitation would firstly, be employed, secondly, followed by struvite crystallization.

Quantitative mineralogical phase analysis of the final products
In agreement with the previous finding of the calcium effect on the struvite crystallization process (Le Corre et al., 2005;Doyle & Parsons, 2002;Wang et al., 2005), many calcium phosphate minerals were suggested to precipitate other than struvite in the MAP solution with calcium at the higher ratio of Ca/Mg of 1. Consequently, multi-minerals would precipitate leading to the reduced quantity of struvite, although recovering phosphorus from the waste stream could be very efficient and productive. While a good slow-release fertilizer reflects the morphology and purity of struvite (Rahman et al., 2026-2030, Liu et al., 2013, the amount of struvite produced could be measured in terms of the economic feasibility of the precipitation method. In the study, the mineralogical composition of precipitates was determined according to the XRPD Rietveld refinement method, and the quantitative results according to the Ca/Mg molar ratios and its solubility product constants of the identified minerals (pKsp = -log 10 Ksp) are given in Table 3. At this point, the percentage by weight (wt. %) of each phase was calculated using the refined lattice parameter of the Rietveld method. Moreover, the pKsp values were used to understand the possibility of mineral precipitation from the thermodynamic perspective, in which these values determine the dissociation compound in water in that the greater the pKsp is the more soluble the compound. Instead of sylvite, all minerals presented in Table 3 with the positive values of pKsp are considered to have a high possibility for precipitation of the solution subject to the present study.
Firstly, in the absence of Ca 2+ , struvite (58.45 wt. %), struvite-(K) (40.21 wt. %) were major minerals found in the collected precipitates with minor sylvite (1.34 wt. %). In this regard, the lower level of ammonium and the higher level of phosphate activity affected the generation of struvite-(K) crystal (Bouropoulos & Koutsoukos, 2000;Song, Donnert et al., 2007). In fact, the proportion of struvite and struvite-(K) products was not significantly modified when the concentration of calcium was less than 20 ppm in the solution. Additionally, the pH solution of 9-10.5 is suggested for optimal phosphate and potassium recovery in struvite and struvite-(K) products (Bouropoulos & Koutsoukos, 2000;Song, Donnert et al., 2007). Nevertheless, the calcium dosage is increased to make the adverse impact becoming pronounced. An increase in calcium concentration in the solution (Ca/Mg ratio of 0.5), made three dominant Ca-bearing minerals, namely dolomite, hydroxyapatite, and portlandite formed, where hydroxyapatite is major (13.01% by weight). Conversely, the reduction in Ca-concentration is more likely to have been initiated largely by the precipitation of hydroxyapatite.
Also, the reduced phosphate concentration in precipitates may correspond to the formation of amorphous calcium phosphate (ACP) as previously suggested by XRPD analysis (Figure 5; Desmidt et al., 2013). When the solution chemistry was set up at a Ca/Mg ratio of 1, struvite and struvite-(K) (7.8 and 20.20 % by weight, respectively) seemed to be inhibited by the formation of newberyite and Mgwhitlockite. Apparently, newberyite crystallized relating to the interaction of magnesium and phosphorous contents in the solution, while its crystallization is usually accompanied by struvite (Abbona et al., 1988). Accordingly, the newberyite forming reaction can be interpreted as follows (Abbona & Boistelle, 1979;Kontrec et al., 2005): As reported previously, newberyite could be favorably formed in the lower pH solution (pH 6; Abbona & Boistelle, 1979). In the current experiments, however, struvite might be formed earlier than newberyite, in that the crystal nuclei of struvite supposedly promoted newberyite in the formation of crystal clusters (Abbona & Boistelle, 1979). In the 1: 1 molar ratio Ca/Mg experiment, Mg-whitlockite, and hydroxyapatite could be generated with the comparable amount as confirmed by the XRPD Rietveld method (Downs & Hall-Wallace, 2003). Those calcium phosphate minerals could be simultaneously crystallized in wastewater at room temperature (Lagier & Baud, 2003), and had been evidently found in the hydrothermal solution (Bayuseno & Schmahl, 2020;Li et al., 2016).
In addition, the corresponding struvite content in calcium experiments with Ca/Mg at 2:1 had disappeared, suggesting no interaction of ammonium and phosphate under this condition (Abbona et al., 1986;Gunay et al., 2008). Unexpectedly, a relatively large amount of sylvite (9-34% by weight) was recovered from the deposits (with Ca/Mg molar ratio < 0.5) that could be obtained during the drying sample. Moreover, the discovery of the minerals Ca-and Mg-phosphate allowed controlling the determination of the calcium content in the MAP solution, which influences the purity and productivity of struvite. This level of Ca concentration should be set in the actual wastewater by quantifying sufficient additional magnesium and potassium ions to compensate for the calcium concentration, which enables struvite and struvite-(K) to be produced at an economic value (Li et al., 2016;Yan & Shih, 2016).

Main findings of the present study
A stirred-batch lab of MAP crystallization under varying Ca/Mg ratio and pH solution was successfully demonstrated in the study, in which the kinetics of crystal growth for struvite and struvite-(K) as phosphate minerals that often generate a tenacious scale of industrial equipment. This study established the kinetic model representing curves of the pH changes versus the Mg 2+ sensitivity under various experimental conditions. The kinetic analysis was focused on calculating rates of mineralization reactions and validating kinetic equations accordingly. Moreover, the crystallization rate of solid phase products could be derived from two reaction stages, including a severe initial pH reduction, accompanied by a gradual pH reduction until the end of the experiments. In the reaction process, the crystallisation kinetics of the precipitates obeyed two stages according to the pH reduction in the time periods of 0-3 min and 3-60 min, whereas both patterns appropriately fit the first-order kinetics.
In the case of minute Ca 2+ amounts (1, 10, and 20 ppm) present in the MAP aqueous solution, faster solid-phase growth rates could be observed in both time periods that might be related to the crystal growth rates of struvite and struvite-(K) crystals, according to the XRPD Rietveld analysis (Table 3). However, growth rates dropped steadily as higher Ca concentrations were added (Ca/Mg molar ratio > 0.5). As a result, this condition increased the number of minerals that decreased growth rates. Additionally, the morphology of solid precipitation products was greatly changed. Therefore, this kinetic relationship provided a theoretical basis for the struvite and struvite-(K) crystal production with desired morphology. In fact, a significant effect on crystal morphology and structure developed during experiments might be contributed by the difference in crystal growth of Ca-and Mg-phosphate-bearing minerals.
Further investigation by the quantitative XRPD Rietveld analysis (Table 3) supported the previous finding in the literature that the competition of Ca-P and Mg-P promoted many minerals (e.g., hydroxyapatite, dolomite, Mg-whitlockite, and Newberyite) grew simultaneously during struvite crystallization (Xiaoning Liu & Wang, 2019;Yan & Shih, 2016), in which interfering of Ca made obstruction of the struvite crystallization. The rationale for the crystal growth for hydroxyapatite found in the study might be related to the nucleation of Ca-P easier than the MgP nucleation because of its lower solubility (Xiaoning Liu & Wang, 2019). For this reason, hydroxyapatite is more likely to precipitate preferentially rather than struvite at a higher Ca/Mg ratio (~2). Correspondingly, an approach of controlling Ca/Mg ratio ions present in the MAP solution should be adopted for keeping a minimum level of these ions to enhance the struvite crystallization (Xiaoning Liu & Wang, 2019).

Conclusion
Struvite and struvite-(K) could be crystallized to recover phosphate, ammonium, and potassium levels in the presence of Ca-ion, which is common in wastewater. In this study, the impact of calcium ions on crystallization kinetics, mineralogical phases, and morphology were examined by analytical techniques. Overall, the crystallization kinetics of minerals from the aqueous solution included two stages according to the pH reduction: a severe initial reduction, accompanied by a gradual reduction until the end, in which growth rates could be derived from equations fit with first-order kinetics. Struvite and struvite-(K) crystallization rates were faster by Ca 2+ ions present at limited doses (from 1 to 20 ppm). At the first stage, however, a gradual increased in the rate constants was observed with increasing Ca doses added to Ca/Mg > 0.5, while the second stages showed the gradual reduction of crystal growth rate. The quantitative XRPD analysis confirmed that lower concentrations of Ca (<20 ppm) did not influence the struvite and struvite-(K) crystallization. However, its adverse impact was evident with the increase Ca/Mg molar ratio to 0.5, which corresponds to the formation of Ca and Mg phosphates. Specifically, hydroxyapatite in terms of amorphous and crystalline phases was proposed to dominantly precipitate from the solution under higher Ca/Mg ratios (>0.5). For samples at Ca/Mg molar ratios of 1 and 0.5, the amount of hydroxyapatite was comparable with the quantity of struvite and struvite (K). but its amount became a major phase at Ca/Mg molar ratio of 2 along with the formation of Mg-whitlockite and newberyite. The struvite crystal with needle morphology could be observed at a higher Ca/Mg ratio, but in the subsequent decrease in Ca/Mg ratio, this morphology changed to an irregular shape. Further studies should focus on the control of calcium and magnesium in solutions at a low proportion for recovering phosphate and potassium from waste streams in order to produce the high quality and quantity of struvite and struvite-(K).